Dacia

Dacia examines the way the Roman conquest and organisation of thecentral core of the province of Dacia impacted on the native settlementpattern and society. It analyses Roman-native interaction from a landscapeperspective focusing on the core territory of both Iron Age and RomanDacia. This includes the royal Dacian heartland (the Orastie Mountains)and its surrounding lowlands, which later belonged to the hinterlands ofUlpia Traiana Sarmizegetusa and Apulum, the two most important Romantowns in the province of Dacia.Oltean considers the nature and distribution of settlements of various types

and functions in the pre-Roman and Roman periods, the human impact onthe local landscape and the changes which occurred as a result of Romanoccupation. The study also provides a methodological framework for furtheranalysis of the landscape and of the evolution of the settlement pattern, whichcan be extended throughout the province of Dacia and into neighbouringareas. The result is a detailed consideration of previous theories of nativesettlement patterns and the impact of Roman colonisation. Dacia offers freshinsight into the province of Dacia and the nature of Romanisation andRoman-native interaction.

Ioana A. Oltean is a Post-doctoral Research Fellow at the University ofGlasgow.

DaciaLandscape, colonisation and romanisation

Ioana A. Oltean

First published 2007by Routledge2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN

Simultaneously published in the USA and Canadaby Routledge270 Madison Ave, New York NY 10016

Routledge is an imprint of the Taylor & Francis Group,an informa business

All rights reserved. No part of this book may be reprinted orreproduced or utilised in any form or by any electronic, mechanical,or other means, now known or hereafter invented, includingphotocopying and recording, or in any information storage orretrieval system, without permission in writing from the publishers.

British Library Cataloguing in Publication DataA catalogue record for this book is available from the British Library

Library of Congress Cataloging in Publication Data

ISBN10: 0–415–41252–8 (hbk)ISBN10: 0–203–94583–2 (ebk)

ISBN13: 978–0–415–41252–0 (hbk)ISBN13: 978–0–203–94583–4 (ebk)

This edition published in the Taylor & Francis e-Library, 2007.“To purchase your own copy of this or any of Taylor & Francis or Routledge’scollection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”

ISBN 0-203-94583-2 Master e-book ISBN

© 2007 Ioana A. Oltean

Contents

List of figures viAcknowledgements xi

1 Introduction 1

2 The Dacian heartland 26

3 The historical setting 41

4 Settlement and society in the late pre-Roman Iron Age 60

5 The Roman social landscape 119

6 The romanisation of the landscape 207

Bibliography 228Index 240

Figures1

1.1 Map of Dacia and neighbouring provinces. 21.2 Low-altitude photograph from a kite of a Roman building

visible as a negative cropmark, north of Alba Iulia(V. Barbuta). 14

1.3 Flight track logs in the study area (1998–2004). 151.4 Ulpia Traiana Sarmizegetusa – general view of the Roman

town and modern village from the east, with the Iron Gatesof Transylvania in the background. 20

1.5 Apulum – general view from the south of the site of theColonia Aurelia, now the Partos suburb of Alba Iulia; theeastern, northern and the early western sides of the enclosureare visible on the edges of the modern occupation. 21

1.6 General view of the Gradistea Muncelului hill, the site of theDacian capital, Sarmizegetusa Regia. 22

1.7 Remains of Roman sites as represented on the 1870–1875Austrian cadastral map of Transylvania: coloniaSarmizegetusa and its surroundings (The Austrian StateArchives, Vienna). 23

1.8 The auxiliary fort at Micia as represented on the 1870–1875Austrian cadastral map of Transylvania (The Austrian StateArchives, Vienna). 23

1.9 Line of the Roman road along the Mures from the rivercrossing near Gelmar, towards Sibot (The Austrian StateArchives, Vienna). 24

2.1 Map of the main topographical features and the location ofmodern settlements mentioned in text. 27

2.2 River meanders in the Strei valley near Calan. 30

1 The copyright of aerial photographs, unless specified otherwise, belongs to W. S. Hanson.

List of figures vii

2.3 General view over the Luncanilor Platform in the OrastieMountains showing small, subsistence-based cultivatedfields interspersed with habitation, orchards and pastures.Numerous small platforms/terraces of various dates are alsovisible in the pasture. 34

4.1 Distribution of pre-Roman settlement in central Dacia (thedense occupation along the Orastie Valley is shown in moredetail). 64

4.2 Aerial photograph showing the fortified site at BreteaMuresana largely destroyed by modern quarrying. 65

4.3 Aerial photograph of man-made platforms/terraces in theGradistea Muncelului area, probably late Iron Age in date. 68

4.4 Dacian house plans in upland settlements. 704.5 Aerial photograph of the hillfort at Costesti-Cetatuie. 774.6 Aerial photograph of the hillfort at Costesti-Blidaru. 774.7 Aerial photograph of the Piatra Rosie hill at Luncani; the

small enclosure of the hillfort is located on the cleared topof the hill, while the larger enclosure lies under the trees. 81

4.8 Aerial photograph of the citadel hill at Deva. 824.9 Aerial photograph of the hillfort at Cugir. 824.10 Near vertical aerial photograph of the hillfort at Capalna

and its surroundings; the hillfort is located on the clearedand flattened top of the hill, with one tower visible in thecentre of the photograph. 83

4.11 Aerial photograph of the hillfort on the Sampetru hill,Hunedoara; the large enclosure dates to the Hallstattperiod, but a smaller circular enclosure, visible towards oneend (left), is perhaps related to the Dacian occupation ofthe site attested by artefactual discoveries. 84

4.12 Comparative plans of hillforts indicating their layout,architecture and size (after Gheorghiu 2005; Figures 50, 52,53 and 55). 85

4.13 Plan of the Dacian hillfort at Cigmau. 864.14 Distribution of Dacian hoards in relation to settlement

locations (settlement buffers at 2 kilometres). 924.15 Aerial photograph of a buried multi-period settlement at

Simeria. 964.16 Aerial photograph of the area sacra at Sarmizegetusa Regia. 994.17 Distribution of pre-Roman industrial activity. 1024.18 Ancient limestone quarry along the southern side of Magura

Calanului hill (I. Oltean). 1034.19 Distribution of pre-Roman spiritual activity. 1085.1 Comparative distribution of Dacian and Roman

settlements. 121

viii List of figures

5.2 Distribution of Roman villas and individual homesteads inrelation to towns. 122

5.3 Aerial photograph showing funerary constructions at AlbaIulia (see transcription 5.36 A). 124

5.4 Aerial photograph indicating the buried remains of a singlestone building near Balata. 125

5.5 Examples of excavated villa complexes and buildings incentral Dacia (Hobita, Deva, Manerau, Cincis, Aiudul deSus) and in the north-western Dacia, around Napoca(Chinteni, Ciumafaia, Apahida). Rooms with hypocaustsmarked in grey (Manerau and Apahida not oriented tonorth and the two buildings are not in their correct spatialrelationship). 126

5.6 The site of the villa at Santamaria Orlea from the air;the remains of two to three buildings (some partiallyflooded) are visible in the foreground and furtherpartially extant building remains are visible in thebackground. 127

5.7 Remains of excavated villa buildings at Strei, and theadjacent early medieval church from the air. 128

5.8 Plans of villa and probable villa sites discovered from the air. 1285.9 Buried remains of multiple buildings of the Roman villa at

Oarda. Oblique aerial view from the north-west. 1295.10 Aerial photograph of a multi-period site at Vintu de

Jos; cropmarks indicate the presence of a Dacian andDaco-Roman village of sunken houses and storage pits inthe immediate vicinity of a Roman villa. 130

5.11 Site plan of Vintu de Jos based on aerial photographicevidence. 131

5.12 Aerial photograph of a probable villa site at Micesti, northof Apulum; buried remains of stone walls or foundations areindicated by cropmarks. 132

5.13 Plan of two probable villa sites discovered from the air onDelinestilor (Sucioni) hill at Hobita. 133

5.14 Aerial photograph of a possible villa at Hobita on Sucionihill (2). 134

5.15 Set of millstones from a probable villa at Hobita(Figure 5.14). 135

5.16 Plan of the villa house at Aiudul de Sus showing differencesin wall widths and a possible interpretation of internal spacedivision. 141

5.17 Distribution of Roman aggregated settlements. 1445.18 Aerial photograph indicating the buried remains of a

village with sunken and (possibly) surface-built houses atBerghin. 146

List of figures ix

5.19 Transcription of the archaeological features based on aerialphotographic evidence at Berghin. 146

5.20 Plan of fragmentary remains of multiple stone-buildings ofprobable Roman date on the northern edge of the modernsettlement at Vintu de Jos based on aerial photographicevidence. 149

5.21 Aerial photograph indicating buried remains of multiplestone buildings at Batiz, probably from a Roman village. 150

5.22 Buried remains of one stone building and possible traces ofseveral others south of Benic. 151

5.23 Traces of quarrying and associated settlement of probableRoman date at Cozia; the circular structure (left foreground)could be late prehistoric. 152

5.24 Calan-Bai (Aquae) Roman stone pool and adjacent moderninstallations. 153

5.25 Germisara Roman spa and ritual complex near modernpools at Geoagiu-Bai. 154

5.26 Roman andesite quarry at Uroi (Petris?) in the Mures valley. 1545.27 Excavated area of the military vicus at Micia – thermae and

amphitheatre. 1565.28 Site plan of the Roman fort and military vicus at Vetel-Micia. 1575.29 Site plan of the Roman auxiliary fort and military vicus at

Cigmau-Germisara. 1585.30 Site plan of the Roman fort and military vicus at Razboieni

(Salinae?). 1595.31 Site plan of the Roman town at Sarmizegetusa Ulpia. 1665.32 Aerial photograph of forum 1 at Colonia Sarmizegetusa. 1675.33 Colonia Sarmizegetusa: the amphitheatre, temples and

buildings outside the enclosure near the north gate. 1675.34 General plan of the Roman settlement at Apulum. 1685.35 Plan of structures within the Colonia Aurelia

Apulense (Alba Iulia-Partos) based on aerial photographicevidence. 169

5.36 Plans of the Roman cemetery north of the canabae atApulum (A); and of archaeological features inside theMunicipium Apulense (Alba Iulia) (B) based on aerialphotographic evidence. 171

5.37 Distribution of Roman settlement in relation to major andsmall towns (buffers at 1 kilometre intervals, with 10 and15 kilometres buffers around major towns); topographyindicating areas above 400 and 1,000 metres. 176

5.38 Distribution of Roman buildings or construction materialsin relation to roads (buffer zone at 3 kilometres). 178

5.39 Distribution map of Roman industrial activities in relationto settlement evidence. 182

x List of figures

5.40 Extensive extant remains of gold quarrying activity ofRoman (possibly Dacian) and later date at Pianu de Sus. 183

5.41 Limestone quarry (foreground) to the west of SarmizegetusaUlpia (background left) (I. Oltean). 184

5.42 Distribution of Roman spiritual activity. 1885.43 Traces of a probable Roman cremation cemetery at Vintu

de Jos. 1925.44 Roman terrestrial and riverine transport network. 1945.45 Roman funerary altar re-used as a pilaster inside the early

medieval church at Densus, north of Sarmizegetusa Ulpia. 2026.1 Schematic model of settlement type and hierarchy in the

late Iron Age (above) and the Roman (below) periods. 2086.2 Distribution of Daco-Roman settlement. 2126.3 Distribution of military sites and materials. 218

Acknowledgements

This study would not have been possible in Romania only a few years ago,before the first aerial reconnaissance programme exclusively for archaeolo-gical purposes was started by Prof. W. S. Hanson and myself in WesternTransylvania (1998–2004). Apart from providing the vital photographs andmaps, the project introduced a new perspective to archaeology in the regionwhich has been promoted throughout the present volume. I am very gratefulto Prof. W. S. Hanson for setting up this project, to the financing bodies(the Leverhulme Trust and the British Academy), all the collaborators andthe local authorities that ensured its running. This new approach to thearchaeology of Dacia became apparent throughout the session on RomanDacia organised by Prof. Hanson and Dr. I. Haynes (London) at the RomanArchaeology Conference in Glasgow in 2001. The papers presented werelater published by the organisers in the Supplementary Series of the Journalof Roman Archaeology (Hanson and Haynes 2004).The British Academy has continued to provide great financial support

both towards the implementation of systematic aerial research programmesin Romania and the better understanding of the nature and impact ofRoman colonisation of the Dacian territory. This has allowed the research oncentral Dacia to be further developed by expanding it into new areas of theformer Dacian kingdom as part of my current Post-Doctoral Fellowship andthrough a subsequent Large Research Grant. However, it was the Universityof Glasgow which provided the financial support which first enabled meto undertake this research. The University’s anniversary 2001 PostgraduateResearch Scholarship funded my PhD research in conjunction with theOverseas Research Student Awards scheme. Various internal funds from theGraduate School of Arts and Humanities and from the School of Historyand Archaeology at Glasgow University have also supported the productionof illustrative material and visits to sites on the ground and to relevantmuseums and libraries in Cluj-Napoca, Deva and Alba Iulia. In addition,the Derrick Riley Bursary for Aerial Archaeology supported the acquisitionof Corona satellite imagery.Prof. Hanson provided a high standard of academic excellence in the

supervision of my PhD research that constitutes the basis of this monograph,

xii Acknowledgements

patiently providing his full support for its completion; first, as my researchsupervisor and subsequently as my departmental post-doctoral mentor, hehas greatly influenced the way my own thoughts and ideas have beenexpressed throughout this study. Dr. I. Haynes (Birkbeck College, Universityof London), who was the external examiner for my PhD thesis, has beentremendously supportive throughout my research and has kindly offeredadvice and data relevant to this study arising from his own research projects inApulum and its hinterland. The academic staff in the Department of Archae-ology in Glasgow University, along with my fellow postgraduate studentsduring 2000–2004, have created an extremely challenging and stimulatingenvironment throughout my time here. Prof. I. Piso and Dr. A. Diaconescu(Babes-Bolyai University of Cluj Napoca) provided me with a solid basis ofknowledge of Roman history and archaeology, and guided my first attemptsat academic research.To R. Palmer (Cambridge) for additional instruction in air-photo

interpretation and mapping, useful comments and advice, and for hissense of humour; to him, and to Prof. I. Scollar (Bonn), Dr. M. Doneus,Dr. W. Neubauer (Vienna) and H. Corley (Glasgow) for help in persuadingthe computers to help me in site mapping and GIS design and analysis;also, to Dr. D. Ruscu, Dr. C. Ciongradi (Cluj Napoca) and Aura Popa(Los Angeles) for useful comments and suggestions on parts of my earliermanuscript; to them and to Dr. L. Ruscu, V. Radeanu, Dr. E. Bota, mygratitude for help in gathering the relevant bibliography for this study andfor constant moral support; to Dr. C. Ciongradi and Dr. G. Gheorghiu(Cluj Napoca) who facilitated my access to information from their PhDtheses at the Cologne University and the Babes Bolyai University ofCluj-Napoca respectively; to the support staff of the Department ofArchaeology in Glasgow; last but not least, to my family for all their moralsupport and above all, patience; to everybody, thank you!

Ioana Adina OlteanDecember 2006, Glasgow

1 Introduction

1.1 The state of research on Roman Dacia

One reason for Dacia being somewhat special for Roman studies is its rathershort life within the empire, spanning the second and most of the thirdcentury ad, covering the period when the Roman Principate experiencedboth the peak of its power as well as the onset of decay. Conquered forstrategic as well as political reasons by Trajan in the age of the greatestterritorial extent and power of the empire, it was abandoned later in the thirdcentury ad because of political infighting affecting the empire at its core andthe subsequent failure to control the limes areas. The second and the earlythird centuries ad are generally accepted as the epitome of Roman imperialrule, which continues to serve as a model in various comparisons. In addition,by the end of the first century ad the empire had refined its approach toconquest and further organisation of new provinces. Consequently, studyingthe implementation and impact of Roman administration in Dacia providesa unique insight into the pattern of conquest and occupation of provincialterritory at the height of Rome’s power.One of the great expressions of Roman power in conquered territories

is the impact on the natural and cultural landscape. As argued elsewhere(Oltean 2004), the current orthodoxy concerning the impact of Roman occu-pation on Dacia (Figure 1.1), the implementation of Roman rule and thedevelopment of the provincial landscape is based on a few, seemingly, gener-ally accepted points. First, the Roman conquest is seen as a dramatic event,involving massive colonisation. Several literary sources describe the process;among them Cassius Dio (LXVIII 14, 4), who refers to Trajan’s policy ofcolonisation with specific reference to urbanisation; and Eutropius (VIII 6,2), who mentions significant colonisation from all around the Roman worldand, indeed, a high Roman citizen-presence in Dacia ‘to occupy its landsand cities’ at the beginning of Hadrian’s reign, as a result of a deliberateTrajanic policy to redress the depopulation of Dacia resulting from the longwar against Decebalus. These examples suggest that the phenomenon wasimportant enough to have come to the attention of ancient historians (for themost recent discussion see Ruscu (2004)) and, at least superficially, a number

Figu

re1.

1Map

ofDaciaandneighb

ouring

provinces.

Introduction 3

of archaeological discoveries appear to support these records. Another leit-motif throughout modern literature is that the native Dacian populationwas moved from their hillforts and settled in the lower areas, which wereeasier for the Roman army to keep under control; indeed, the archaeologicalevidence for a violent destruction of the major Dacian hillforts (which do notseem to continue to be in use during the Roman period) has been interpretedas occurring within the chronological context of the conquest (Glodariu1993, 15). Furthermore, it is generally asserted that the Roman policy ofcolonisation was rather brutal, with the authorities depriving the natives oftheir lands for town foundations, for colonists and veteran’s land holdingsthat established villa estates, for army needs and for imperial estates (Protase1968, 510). Finally (and somewhat surprisingly given the above theories),it is widely believed that the natives lived in harmony with the colonistsand romanised themselves during the two centuries of Roman occupation,and that together these two groups continued to settle these lands after theRoman administration and military forces left Dacia at the end of thirdcentury ad (see Protase 1980, 228–52).But before analysing the archaeological basis of this theory in more detail,

it is necessary to review the current state of archaeological research andarchaeological landscape in Dacia. Despite the efforts made during the twen-tieth century, archaeological research has tended to concentrate on military(e.g. Gudea 1997) and urban sites, the latter restricted predominantly tosites with proven municipal or colonial status. A number of them attractedinterest from antiquarian and early academic research, but after the cessationof excavations at Sarmizegetusa Ulpia in 1938 and until that work recom-menced in 1973, the archaeology of Roman towns in Dacia was limited tooccasional excavations at Romula and some rescue work at Apulum, Drobetaand Napoca. In the meantime, modern development destroyed the sites ofthe municipia at Dierna (1968–1969) and Ampelum (1985–1986) withoutappropriate rescue work being undertaken and published. Research has beenre-launched and intensified, particularly in the last two decades, with anumber of research projects targeting a number of objectives at SarmizegetusaUlpia, Apulum, Napoca and Tibiscum, some of them involving collabora-tions with French, German and British archaeologists (Diaconescu 2004b).However, a significant number of sites which, as far as we are aware, didnot achieve municipal status, but which are accepted elsewhere as havingat least a semi-urban function (e.g. Burnham and Wacher 1990) have beensomehow neglected; only a few military vici, for example, have been subjectto any excavation (e.g. Tibiscum, Casei, Porolissum, Micia).Rural settlements have been approached only sporadically and with a low

priority. The first decades of the twentieth century saw the beginning of theconsideration of rural settlements, with excavations at several villa sites, manyof them extant at that time, such as Manerau in 1912, Apahida and Garbouin 1913 (see Mitrofan 1973, 127–50, with full bibliography for the firstpublication of these early excavations). A second period of revival of interest

4 Introduction

was noted in the 1950s and 1960s, though for political reasons linked to theagenda of the communist regime. Excavations were conducted on villages(vici) or the cemeteries belonging to them (e.g. Obreja, Soporu de Campie,Bratei, Cristesti, Micasasa – see Protase 1998; Protase 1980, 38–85 withbibliography; Husar and Man 1998; Mitrofan 1999). Similarly, several villasites were also researched, such as Hobita, Deva, Santamaria Orlea, Aiud,Cincis, Chinteni (Floca 1953; Floca and Valea 1965; Winkler et al. 1968;Popa 1972; Mitrofan 1973, 1974, 1976; Alicu 1994, 1998).In recent decades, efforts have been made to record all archaeological sites

within the territory, improving the older data with new information obtainedthrough field walking. Unfortunately, an ambitious project to produce ageneral archaeological gazetteer of Romania was never completed, thoughin some cases it resulted in the publication of regional gazetteers of severalcounties or geographical areas. For most of the discoveries, however, theinformation is scattered in studies at various levels of comprehensivenesspublished in various Romanian archaeological periodicals. Despite its badreviews (Daicoviciu 1969), Tudor’s book Orase, tirguri si sate in DaciaRomana (1968) used to be the largest collection of published informationon Roman settlements in Dacia that specialists could rely upon. But nowafter more than 30 years, the information needs to be updated, and the sameapplies to the Tabula Imperii Romani (L-34, Budapest and L-35, Bucharest)whose information continues to be used by the editors of historical atlases(e.g Talbert 2000). This will hopefully be redressed in the future, throughthe efforts of the Institute for Cultural Memory in Bucharest (cIMeC) tocreate a large database of the archaeological sites of Romania accessible onthe Internet to scholars internationally, through a European Union fundedproject (http://archweb.cimec.ro).

1.2 Biases and limitations of current research

As noted above, the immediately recognisable bias in academic research onRoman Dacia is the imbalanced focus on military and urban sites that haveattracted the limited number of specialists and the funds available (Alicu1998, 127–8). Without being a problem particularly related to Romanian oreven to Roman archaeology generally, this alone is a significant bias inducedin the archaeological evidence. Yet in the archaeology of Roman Dacia, thereare numerous biases that apply not only to the quantity, but to the qualityof current information and, since one bias can be a direct consequence ofanother, the end result is that the theories generated from the data cannotbe other than ill-founded.

1.2.1 The influence of history and politics on archaeological research

Until now, archaeological research in Romania has been subservient to estab-lished historical theories generated by literary sources. This attitude is deeply

Introduction 5

rooted in concepts of the past held by modern individuals, among them wereprofessional scholars. Although the ultimate goal of archaeological researchis naturally directed towards explaining historical evolution and phenomena,the visible tendency over the past century has been to rely primarily onthe existing literary sources, despite the fact that their inherent biases couldpotentially be reduced by reference to the totality of the evidence. A generalproblem in the study of the Roman Empire is that archaeological evidencehas been considered only when it supported the historians’ arguments, ratherthan attempting to build a theory based on both sources of evidence. Thedanger is that in such a situation, the literary sources alone tend to generatethe conclusion. More recent comprehensive studies have tried to address thisissue. This situation might be explained sometimes by the paucity of archae-ological information that still applies to some extent in Dacia. Unfortunately,even where that evidence is available, other factors distort its considerationand the resulting conclusions.There is nothing new in the recognition of the importance that historical

models have for political discourse in general, but for a long time in Romaniahistory itself was entirely subordinated to politics. As a result, various subjectsof archaeological research were approached and funded only when and ifthey were seen to serve political fashion trends. After the Second WorldWar, politics became more intrusively and, indeed, aggressively involved indifferent aspects of research as with life in general. Above all, the way ofthinking and writing history had to be Marxist (based on the theory ofhistorical materialism) and no other approach was perceived as ‘suitable’.Ever since 1947, and especially in the 1950s and 1960s during Gheorgiu-Dejrule, the key role of historical research was to feed the discourse of communisttheories regarding the social classes’ antagonisms and the rejection of westernimperialism. In this context, research at the major Roman sites, includingUlpia Traiana Sarmizegetusa, the provincial capital, ceased to be financed(Diaconescu 2004b, 88), and both funds and personnel were re-directedto undertake research on native settlements of pre-Roman, Roman or post-Roman date. But, despite the benefit of re-directing research towards sitespoorly approached before, the conclusions of the research had to fit thegeneral theory of the oppression experienced by the native masses under theimperialist occupation. This idea, born in the middle of the industrial era,had to be sustained by history and archaeology and proven to have existedfor a long time, along with social antagonisms. The Roman conquest andoccupation of Dacia was described as a negative event, not only in relationto the treatment applied to the natives, but also because of the introductioninto the conquered territory of a socio-economic system based on slavesand the exploitation of the lower classes (Constantinescu et al. 1970, 53).According to these scholars, who were projecting ideas usually connectedwith the concept of modern nations into the context of an ancient society,the natives lost their liberty as a people and their properties in favour oftheir Roman conquerors. The opinion expressed by Macrea (1969, 457)

6 Introduction

was that ‘the Roman conquest brought for Dacia not just economic, socialand cultural progress, but also an entire cortege of suffering, pillage andunmerciful exploitation for the slaves and the lowest strata, all inherent to anantagonistic society based on slave exploitation’ (my translation). Similarly,the authors of ‘Inscriptiile Daciei Romane’ (Russu 1975, 22) wrote in theirhistorical introduction: ‘The Roman occupation brought […] real progressin modes of organisation, material and spiritual culture, technology andproduction, and in the way of life; but it was at the same time a regimeof pillage and rough exploitation of the lower, working classes of societyand especially of the masses of native subjugated inhabitants and workers,expropriated in their own country’ (my translation).The other main political and historical theme was the concern to produce

historical arguments to support the political claims over Transylvania enter-tained by, on the one side, Romanian and, on the other, Austro-Hungarian,then Hungarian political circles. Dacia and its inheritance has been a centraltheme in defining national identities of the modern era in the area and it isthe subject of a long-term debate in Romanian and Hungarian historiography(the Romanian point of view is mainly summarised in Daicoviciu et al.(1963), while the Hungarian thesis is presented most recently by Vékony2000). During the communist years, the subject was particularly in fashionin the 1970s–1980s, during the rule of N. Ceausescu, when communistpropaganda in Romania gained a strong nationalist message. Within thiscontext, one of the main concerns of Romanian historians was to bringforward arguments for the rapid and durable romanisation of the natives inDacia as a major element in the context of the ‘theory of Dacian–Romancontinuity’. This thesis was developed over the past three centuries mainlyas a response to the corresponding Hungarian propaganda which attemptsto sustain its political position towards Transylvania and to prove that theterritory was devoid of any population on the arrival of the Hungarian tribes.It is argued that the Romanians emerged in the region through a massivemovement of population from south of the Danube later on, because theDacians had disappeared as a result of the wars of ad 101–102 and ad 105–106 and the Romans had withdrawn all the population in the third centuryad . Therefore, Romanian scholars focused on disproving this thesis, bringingforward arguments to support the romanisation of the native Dacians underRoman rule and the continuity of life in Transylvania from prehistory tothe Middle Ages.The currently accepted theories on the Roman conquest and rule of Dacia

and, most of all, on the romanisation process, had to be fitted in accordingly,in spite of several flagrant contradictions that I wish to address briefly here.The archaeological evidence so far attests the presence of colonising elementsin both urban and rural contexts. But if in the urban andmilitary sites funerary,epigraphic and other types of evidence seems to indicate quite a large popula-tionwith an origin other than indigenous (see alsoCiongradi 2004a and b), thecurrent level of archaeological data for the rural areas of Roman Dacia makes

Introduction 7

it impossible at this point to assert a similar percentage of colonised elementsthere. The destruction of the Dacian hillforts and the population movementhave been accepted with little if any surprise, given the huge effort made by theRomans to conquer these sites. But on a closer examination, current theories onthenature of the conquest and the treatment applied to thenatives are in blatantcontradiction.One of them suggests that the natives lived on the tops ofmoun-tains and were forced to settle in the fertile lowlands, while another states thatthe Romans came and took the fertile lands for their own properties and forcedthe natives tomove away or work on their properties as cheap labour. Unfortu-nately, it is often forgotten that the sole reason for the existence of the hillfortsis as elite sites and the purpose of their location is strategic and status-related.Accordingly, after themilitary defeat and the introductionofRoman rule, therewas no longer a reason for such sites to exist. The extrapolation of this modelof hillfort destruction to the entirety of the Dacian settlement pattern not onlypushes this interpretation to an unsubstantiated extreme, but also would havebeen a questionable policy decision on the part of the Romans. The idea thatthe Roman colonists deprived all the natives of their lands does not sit well withthe benefits of a rapid and durable romanisation through close and peacefulrelationships between all the inhabitants of the new province, whether nativesor colonists, civilians or soldiers. If true, the resulting attitude of the Dacianstowards their conquerors is likely tohavebeen resistance to acculturation, ratherthan receptiveness. It certainly does not take into consideration other factors –cultural or economic, for example – which must have influenced attitudesand the whole process needs a more refined and detailed interpretation.Therefore, it is no surprise that the theory has been challenged in the 1990s

not just by the Hungarian colleagues, but also by some Romanian scholars,especially archaeologists. It is not my intention here to analyse this polemicin more detail, but to show that it has influenced the perception even ofthe existing evidence and in the end it has distorted the general view ofRoman Dacia in terms of conquest, colonisation, administration and, indeed,romanisation. The eventual reaction of some Romanian archaeologists tothis theory is an attempt to evaluate the existing archaeological evidence atits true value, without dismissing the elements that might offer a differentpicture than the one desired by political discourse. It would be wrong toaccept any form of political interference in either archaeological or historicalpractice any longer. The political attempts to influence interpretations of thepast usually reflect a failure to find solutions to present issues (a distractingfactor from real political issues). Moreover, adding a supplementary biasto those that currently apply to archaeological research would distort ourperception of the past even more.

1.2.2 Archaeological information: reliability and accessibility

Reliable archaeological evidence is surprisingly limited. This statement mightsound odd given the amount of data recorded so far for Roman Dacia,

8 Introduction

but unfortunately for the most part the information provides only thegeneral location of archaeological discoveries. Many efforts have been madeto identify and systematically research the archaeological sites of Transylvaniain the twentieth century and in some cases even earlier, especially throughnon-systematic field walking and excavations at various scales. Both TabulaImperii Romani L-34 (Budapest) and L-35 (Bucharest), along with Tudor’sbook Orase, tirguri si sate in Dacia Romana (1968) and the several countyarchaeological gazetteers, refer to a significant number of sites. But the vastmajority of these sites have somewhat stereotypical descriptions, most ofthe time mentioning the ubiquitous ceramic fragments and/or constructionmaterials, when, in fact, these stereotypes relate to the methods of collection,interpretation and, indeed, evaluation of the data for the whole territory. Inaddition, there is still a lack of systematic coverage of the territory and nounitary method of recording. Only some 10 per cent of reported sites havebeen the subject of more extensive excavation projects; the remaining 90per cent are just indicated by finds (artefacts or building materials scatteredon the ground surface). There are no site plans available for most of them,because of a failure to apply modern techniques of site prospection andbecause of restrictions on access to maps. As a result, however, the size andsignificance of these sites has not been fully appreciated and several categoriesof sites characteristic of a landscape are still overlooked. This could be thepotential explanation, for instance, for the fact that no Iron Age or Romanland-use systems have been yet documented in Dacia.But even the information available is extremely difficult to consider and

evaluate as a whole in the absence of an accessible national system of record ofarchaeological sites. Dealing with a type of information that is, by its nature,accumulated over a long period of time, naturally brings problems of storageand accessibility of data. In Romania, the system of publication of the resultsis not very helpful in terms of access. Tabula Imperii Romani L-34 and L-35,and Tudor (1968) are now well out of date. Until a database containingarchaeological reports made since 1983 became available online, along witha basic map distribution (http://www.cimec.ro/scripts/mapserv.exe?map=/mapserver/mapserver_ro.map&mode=browse), more recent discoveries wereaccessed primarily through sporadic, random publication in variousRomanian periodicals. The material might have been addressed in a fewcases in more general studies, which aim to collect the data discovered overa wider area, usually in terms of historical geography, chronology or specificcategories of sites (e.g. Wollmann 1996 for mining and quarrying; Popa1987 for Tara Hategului). There are also few cases of modern regionalarchaeological gazetteers (e.g. Alba County – Moga and Ciugudean 1995).There are no regularly updated archaeological databases. The latest

published collection of Roman rural settlement in Dacia (Popa 2002), evenwithout providing much data analysis, is a useful gazetteer collected overa long time. But, given the lack of databases available, it was deemed tobe incomplete, with sites, for example the marble quarry and settlement at

Introduction 9

Bucova, being left out. The recent efforts to complete a general archaeolo-gical database (by cIMeC – p. 4) are extremely important and the value ofsuch action has been proved already by the availability of information onlineunder the format of a searchable database from excavations from the years1983–1997, 1999 and 2000. The use of the Internet for information, aswell as the availability of digital data to support computerised quantitativeand settlement pattern analysis, will provide a valuable support for futureresearch. Unfortunately, even this latest attempt to preserve archaeologicalinformation does not include the precise geographical locations of the sites,continuing to use location descriptions by place-names, which have alreadyproved to be inadequate.

1.2.3 The problems pertaining to site location

The experience of the last 100 years shows the importance of the accuratetransmission of information in the context of successive changes in archae-ological methodological requirements or even in the territorial adminis-tration system and place-names. The failure to locate archaeological sitesby their geographical co-ordinates and reliance on place-names producessignificant difficulties in attempting to locate some sites that have beenpreviously reported. This occurs especially with place-names of very localsignificance within the area of a particular village, for example, which arenot in use anymore, nor traceable through archive maps or documents. Asa direct consequence of this failure to locate archaeological sites precisely,some of them are very imprecisely located when referred to by variousauthors.This confusion persists even in the most recent publications. For example,

N. Gudea in his study on the Roman military camps in Dacia (1997, 101–2)locates the Roman fort of Cigmau (Germisara) and its civilian settlementapproximately one kilometre to the N of its true position (Hanson andOltean 2002, 114). Unfortunately, this error persists and has the potentialto bias later studies in that area. The process of alteration of information istraceable for example in Benea’s (2000) article on military vici from Dacia.Acknowledging the difficulty of access to information, she tried to assembleall the data available for civilian settlements outside Roman auxiliary fortsin Dacia, and thus produce a useful tool for both Romanian and interna-tional archaeologists. Unfortunately a typing error misplaced the locationof the building complexes known from rescue excavations at Vetel (Micia)published by Marghitan (1970a) by some 250 metres to the east. Theseexamples highlight the difficulty of assembling data, especially for thosesites where an overall site plan has not been produced, and the import-ance of such plans for the subsequent production of accurate archaeologicalmaps of larger territories to support landscape studies and settlement patternanalysis.

10 Introduction

1.2.4 Excavation methodology

The excavation methods applied to the Roman sites of Dacia over the lastcentury have varied. As shown by Condurachi and Daicoviciu ‘after thesecond world war [� � �], the distinctive features of Romanian archaeologicalmethod were the absolute priority it gave to stratigraphy and its preferencefor the exhaustive excavation of large sites to the maximum extent that thecircumstances permitted’ (1971, 20). Unfortunately these principles werenot applied to Roman sites until much later. This fact is well illustrated bythe list that they give of sites where modern methods had been applied bythe date of their publication, which contains not a single Roman example,and can be further confirmed by consulting other excavation reports.Thus, for some of the excavations, especially of villa sites, the published

results give little indication of construction phases. This is most unusual,especially since repairs or changes of plan within buildings, or even changesof use of buildings, are frequently recorded in civilian archaeological contextselsewhere. At Hobita, despite the fact that it is so far the only exampleof a villa site to have the whole enclosed area delimited and its internalarrangements defined, there is no indication of any phases of construction orrepair (Floca 1953). Published excavation reports rarely express any concernabout their failure to identify earlier phases of archaeological complexes. Inthe case of at least two civilian buildings (no. 1 and no. 3) in the vicusexcavated by Marghitan at Micia some 30 years ago (1970a), earlier timberphases were revealed by excavation. Despite the fact that there were otherindications, such as the existence of artefacts, including construction materials(tiles), within the filling layers under the floors, these discoveries occurredonly incidentally while the excavators were trying to reach the deepest levelof the stone wall foundations when the earlier construction trenches forthe timber walls intersected their trenches. However, the excavators madeno attempt to establish the extent and layout of these features, to considervariations of plan or internal space division from one phase to another, orto make a study of the quantitative and qualitative or stylistic evolution ofthe different categories of finds. Excavations such as these have, therefore,produced incomplete site plans where chronological developments are nowimpossible to pursue. Similarly, in complexes where the stone phase wentthrough successive transformations or repairs, these are apparent neither inthe reports nor in the site plans, if available. This issue will be considered inmore depth later, when dealing with specific classes of site, but it is worthstating for the moment that in such cases the interpretation of the internalarrangements is almost impossible to establish accurately.

1.2.5 Archaeological prospection and Dacia

‘Excavation is still synonymous with archaeology in many countries’ (Bewleyand Raczkowski 2002b, 3) and that has been very much the case in Romania

Introduction 11

until very recently. Although excavation gives the most precise and detailedinformation about archaeological sites, not every site can be excavated andthe high costs involved or the amount of time necessary to complete andpublish an excavation are only part of the reason. The main problem withexcavation as a research method is that, by its nature, it damages the siteitself in the process and the areas affected by excavation will no longerrepresent an intact testimony of the past. Moreover, excavation standardsevolve over time, and even top standards at one moment in time will beconsidered insufficient or inappropriate in the future. The modern approachto archaeological heritage conservation is focused on non-invasive methodsof research, and most archaeological sites tend to be excavated only whenthey are endangered by development. Within this context, modern methodsof archaeological prospection have developed as a necessity, both for under-standing and monitoring archaeological sites, but also for prior evaluation ofthe site and its potential as accurately as possible. Most importantly, they offer(aerial photography in particular) the possibility of providing better under-standing of sites from a landscape perspective, and indicating the way thathuman settlements interacted with the natural landscape and with each other.Traditional field walking, that is the recovery of artefacts brought to the

surface during ploughing, is the only method of archaeological prospectionwhich has been applied historically in the study of Roman Dacia (p. 4).Unfortunately, in its application no attention has been given to the unitaryplanning of the field coverage or to systematic data collection and analysis.More recently, the Apulum Hinterland Project international team has startedsystematic field walking in the vicinity of Alba Iulia in collaboration withthe University of Alba Iulia. Field walking alone can locate a site, providesome indication of its extent and, from interpretation of finds, evaluate itschronology. But even at its best, field walking can give only limited cluesas to the nature of the site. What can make the difference in settlementpattern analyses is understanding the site’s full extent and morphology.This would allow differentiation between, for example, an individualhomestead (farm) and a nucleated settlement. In the case of individualhomesteads, the layout of the internal buildings, their individual plan orthe building techniques can distinguish between a villa site and a nativefarm.Geophysical survey has only recently started to be applied in Dacia.

The usual lack of funds, trained specialists or surveying equipmentprobably constitute the main reasons for this, as for aerial reconnais-sance. As in the case of excavations, the first attempts at geophys-ical survey have focused on urban and military sites. Some attemptsat geophysical surveying combining magnetometry and resistivity havebeen made at a number of sites (e.g. Porolissum, Scurtu 1997; Cigmăuhttp://www.cimec.ro/mapserver/asp_script/cronica/detaliu.asp?k=1205 visi-ted 6 December 2006). At Apulum, more extensive areas in the coloniaand the municipium were recently subjected to geophysical survey by

12 Introduction

the international team operating in Apulum and its hinterland (Hayneset al. forthcoming) and extended into the territory around the Romanconurbation with surveys at a number of villa sites (e.g. Oarda-Sesul Orzii http://www.cimec.ro/mapserver/asp_script/cronica/detaliu.asp?k=3182 visited 6 December 2006). However, geophysical survey, whetherutilising magnetometry, resistivity or ground penetrating radar, is by itsnature largely confined to the limits of individual sites. Despite its recognisedvalue in recording details of site layout, it offers only limited opportunitiesto evaluate the site from a landscape perspective and consider other possibleadjacent features that might be related.The aerial view gives human perception a broader perspective. Archaeolo-

gical sites can be recognised even when their degree of preservation is verypoor, whether still visible to some extent on the ground surface or even totallyburied. For more than 60 years it has been proved on numerous occasionsthat, given suitable soil conditions, buried archaeological features can berecognised from the air as cropmarks (Wilson 2000, 16–23; Bewley 2002).This has made aerial reconnaissance extremely valuable, especially for theidentification of previously unknown archaeological features. Furthermore,a trained interpreter can acquire considerable information about a site, bothin terms of its morphology and its probable date, through analogies withsimilar sites whose chronology has been established by other methods (Wilson2000, 65–7, 84–7). In addition, the speed of coverage and consequentlyof analysis of even large territories is significantly higher than through fieldwalking, or indeed geophysical methods. Both these characteristics makeaerial photography the preferred method of archaeological prospection inEurope, especially for landscape research and management. These advantageshave determined the initiation of programmes involving aerial reconnais-sance to acquire new imagery and evaluation of available images from aerialphotographic archives in several countries of Europe at a national scale andon a permanent basis (see Bewley and Raczkowski 2002b, Figure 1). Mostrecently, satellite, multi- and hyper-spectral imagery, or airborne scanningtechniques have been addressed too in an attempt to widen the coverage andaddress the geographical biases in aerial reconnaissance.Probably the first aerial survey and photography of an archaeological site

in Romania was taken as early as March 1918, when Carl Schuchhardttook aerial photographs of the late Roman and Bzyantine frontier wallsin Dobrudja. Despite the fact that these photographs remained unpub-lished until 1954 (Crawford 1954, 208 and Plate VI), Schuchhardt usedthem to correct his own published map of the wall. Unfortunately, thisremained very much the only example of its kind for a long time. Theonly aerial photographic survey programmes over the following decadeswere made for military reasons during the World Wars and the ColdWar, or for civilian mapping purpose, and access to the aerial photo-graphic archives for archaeologists in Romania still remains extremelylimited.

Introduction 13

There have been a few attempts to make use of the available aerialphotographic information in archaeological studies and in the occasionalpublication of sites, but without a clearly structured method of approach.Unfortunately, in all of these cases the information available was insuf-ficiently exploited because of the limited training and expertise of thesepioneers, who were either archaeologists with very little or no experience inmapping or photo-interpretation, or cartographers with photo-interpretationexperience but without archaeological expertise. A group of cartographersfrom the Cartographic Institute in Bucharest published short articlesutilising principles of photo-interpretation to identify possible archaeologicalsites on vertical photographs at Sanislau (Satu Mare) and Dersida (Salaj)in north-western Romania, or at Sarighiol de Deal, Satu Nou and Isacceaalong the Danube in the south-eastern Romania (Rada et al. 1989; Radaand Cochina 1984; Rada et al. 1986). They provide interpretation andsome limited mapping of the archaeological features, unfortunately notalways correct; nevertheless, their attempt to identify previously unknownarchaeological sites is notable. I.O.M. Bogdan-Cataniciu (1981) is oneof the very few archaeologists who gained some access to the Romanianarchives. In her study of the Roman defences of Dacia (Bogdan-Cataniciu1981), she published some examples of extant forts and fortlets visible onvertical photographs. But, as was the case with the material published bythe Bucharest cartographic team, the quality of the reproductions or insome cases even of the original photographs is so poor that the reader mustrely for the most part on the interpretations provided by the author. Farbetter quality of image reproduction is evident in Stefan’s (1986) overviewof known archaeological sites from the air, some of his photographsbeing provided with transcription sketches, but this work has been verypoorly publicised in Romania. Archaeologists have used aerial photographssporadically to illustrate lectures (e.g. C. Craciun mentioned in Ardevan1998, 76), publications of sites (e.g. Tamba 1997, Plate 8; Alicu 1998,Plate 3) or more frequently, to illustrate exhibitions.In Romania, for years, flying for archaeological purpose has been

constrained not only by the lack of financial resources to sustain aerialreconnaissance programmes, or a shortage of specialists experienced in theinterpretation of aerial photographs, but mainly by legal difficulties createdby the restrictions on civilian air traffic in force during the Communistyears (see Braasch 2002; Oltean 2002; Hanson and Oltean 2002, 2003).As a result, all the flight initiatives were no more than sporadic. Stefan’stemporary collaboration with existing cartographic institutions failed tocreate a stronger impact on the real implementation of a similar system ona national scale. Only since the 1990s could E. Pescaru undertake occa-sional limited flights photographing known sites in Hunedoara County fromthe air for illustration purposes, using a helicopter as an aerial platform.Less-fortunate enthusiasts, such as V. Barbuta, were constrained by lack offunds to making photographs using kites as aerial platforms (Figure 1.2),

14 Introduction

Figure 1.2 Low-altitude photograph from a kite of a Roman building visible as anegative cropmark, north of Alba Iulia (V. Barbuta).

with some success especially for the identification of the temple of LiberPater at Apulum (information Dr A. Diaconescu).Unfortunately, even the limited interest in the aerial photographic evid-

ence so far has been almost entirely restricted to sites already identifiedthrough other means or to extant features, easily detectable on the ground.This reduced greatly the potential of aerial photography as a prospectionmethod. Cropmark methodology was effectively ignored. Indeed, until thelast decade, Central and Eastern Europe was thought not to be sufficientlyresponsive to crop mark formation because of the heavy, alluvial topsoil thatcovers the arable regions, combined with a less contrasting deeper geologicalbackground (Bradford 1957, 15 and 23). Moreover, the recognition of cropmarks would have been more difficult from archive vertical imagery. Apartfrom the variable quality of the photographs and their scale, such data islikely to be of more limited archaeological value because the images wereobtained originally for purposes other than archaeology and the very specificconditions necessary to record many archaeological features may not havebeen in force. In 1998, the University of Glasgow started a program ofsystematic aerial reconnaissance in western Transylvania undertaken by Prof.W.S. Hanson, funded by the Leverhulme Trust and subsequently by theBritish Academy (Figure 1.3). The work was undertaken in co-operationwith the National Museum of Transylvania in Cluj, the Museum of Dacianand Roman Civilisations in Deva and the University of Alba Iulia. Thepurpose of the project was to establish the parameters for the application

Introduction 15

Figure 1.3 Flight track logs in the study area (1998–2004).

of aerial reconnaissance in the different environmental, soil and agriculturalconditions pertaining in Romania and also to increase understanding of thehistory and development of the landscape of the area, particularly from laterprehistory to the immediate post-Roman period. The geographical focus ofthe project was south-western Transylvania, particularly the middle Muresriver valley and the plain of Hateg (Tara Hategului) to the south, whichlies at the heart of both the Iron Age kingdom and the subsequent Romanprovince of Dacia. This project has established the first relational databaseof aerial archaeological sites of different dates in Romania and has providedvaluable information for this monograph (pp. 20–21).However, it is generally recognised that the best archaeological results are

obtained through a combination of these various methods of survey: aerial,

16 Introduction

geophysical and traditional. Hopefully, future financing will finally allowsuch a programme.

1.3 Aims and approaches: landscape and romanisation

The relevance of ancient landscape studies to finding the answer to general,theoretical or even more focused issues of the ancient world has been broadlyrecognised. Data interpretation has always pre-supposed assessment of thespatial distribution of similar sites, but understanding of the general patternsor, indeed, unique features revealed by the sites needed to be consideredfrom a broader perspective. Landscape means more than physical space, asdefined by ‘the relationship between humans and their environment, [� � �]society and space’ (Cosgrove 1985, 46). The physical space accommodatinghuman actions suffers a continuous evolution. From an environmentalistview, this would be regarded as a process of degradation resulting from acumulative effect of to some extent natural, but mostly human-generatederosion (Delano-Smith 1996). Amongst all species, humans have been themost successful in using the natural habitat, whether as optimal foragersor as economic men (Ingold 2000, 38). Moreover, humans have adaptedthat habitat to create the landscape, using its resources and transforming itaccording to their necessities. From this point of view, the landscape bearsthe marks of the humans that have inhabited it through time, reflecting theirneeds and way of life, or their level of technological knowledge. Introdu-cing his ‘dwelling perspective’, Ingold (1993, 52) remarks that ‘landscape isconstituted as an enduring record of – and testimony to – the lives and worksof past generations who have dwelt within it and in so doing, have left theresomething of themselves’. The natural environment has brought into theequation factors such as topography and climate, the availability of resourcesfor life (either beneath the earth or on its surface, as local flora and fauna)or, indeed, the defensive potential of particular locations. To this, humanshave added their own leading mindset, needs and technological resources,knowledge, social regulations and politics, which have impacted on the useand, therefore, re-shaping of the landscape. In this way, the evolution of thelandscape is able to tell its story, that of the cultural evolution of humancommunities through time.The approaches presented above define the landscape as a product of

the subjective transformation by man of the objective reality of space. But,the interaction between humans and the environment eventually resultedin a transformation of both. This process would have left significant traces,identifiable by modern archaeological methods. However, the number ofhumans that were accommodated within the landscape at each moment intime would have determined the amount of physical space affected. Analysingthe traces left should indicate more than just the ecological impact of pasthuman activity; it will reveal clues as to the size of that group, or even ofthe way they carried out their transforming activities.

Introduction 17

Going beyond the physical boundaries of each site, the study of settle-ment patterns is far from being solely a mapping or locational process.It involves analysing the spatial and functional relationships of contem-porary sites within particular cultures (Knapp 1997, 5), not solely in a stricteconomic sense, but also in political, social, religious or cultural terms (Knappand Ashmore 1999; van Dommelen 1999). The evolution of landscapesoffers a different understanding of those ancient Mediterranean civilisationsthat were largely town-based, such as those of Greece and Rome, both intheir Mediterranean heartland (Shipley 1996, 8) and beyond (e.g. Dark andDark 1997). But apart from the obvious effect of revealing what provin-cial settlements would have looked like, the study of the settlement patternwithin Roman provinces can address more general issues. The decision tosettle and use a particular space has been taken by people, in groups or asindividuals, in direct relation to their interests. Consequently, by studyingthe resulting impact of their action on the landscape, one should be able totell whether the original effort involved was made by several individuals orby an organised group following a certain policy. Therefore, the nature ofthe colonisation process can be analysed from the way the new Roman-typesettlements emerged within the provincial territory and their effect on theprevious native pattern. According to the current orthodoxy, after the Romanconquest Dacia experienced the first large influx of populations from outsideits cultural boundaries, a phenomenon described by ancient historians andre-enforced by the epigraphic evidence. These newcomers, mostly from otherparts of the Roman world rather than Rome itself, whether granted Romancitizenship or not, had to be accommodated within Dacian territory, asdid the manifestations of the new legal and administrative system and themilitary. The native settlers simply had to comply with the situation.Subject of debate for decades, the approach to romanisation has been

marked by several successive theoretical trends (materialism, colonialism,post-colonialism), all trying to find a satisfactory explanation for what it is anextremely complex socio-cultural phenomenon. Traditionally, the two partiesinvolved, the natives and their conquerors, have been presented as facingeach other from different, sometimes even conflicting or antagonistic posi-tions, reflecting modern political (national) thought on ancient societies. Butromanisation stubbornly gives still inexplicably different, even contradictorypictures, not just when subjected to different theoretical approaches, butalso when seen from different corners of the Roman Empire. Indeed, oneof the main questions in defining romanisation as a process, whether theRomans romanised the provinces (e.g. Garnsey and Whittaker 1978) or,on the contrary, the natives romanised themselves (e.g. Millett 1990), ifanswered at all, finds different responses. On the one hand, this makes aglobal understanding of the process very difficult and, therefore, researchon romanisation has tended to remain at a level of local studies. On theother, it has resulted in extreme attitudes and sometimes its nature or veryexistence has been challenged, either in particular regions (e.g. Africa – see

18 Introduction

Bénabou 1976; Dacia – e.g. Vékony 2000) or as a concept (e.g. Hingley1996; Barrett 1997). Considerable, but so far unconvincing effort has beenspent, therefore, on finding alternative labels (e.g. Webster 2001) to betterdescribe what was essentially a phenomenon of change, occurring as a resultof Roman occupation of new territories. Woolf (1995) rejects ideas likeconflict, competition or interaction in relation to this subject and advances anew interpretative framework, with the creation of a new imperial culture asa structured system of differences replacing both previous cultures (Romanand native), its spread comparable with the growth of an organism that meta-bolises matter. A similar approach to romanisation as ‘a largely consciousprocess by which sections of the indigenous population sought to emulateRoman culture, at least in the form in which they experienced it, motiv-ated by the need to establish their own social status and directly assisted bythe Roman authorities’ was earlier employed by Hanson (1994) in his owninterpretation of the phenomenon in a British context, though positioningit at one extreme of the process of cultural interaction and mutual changewhich occurs whenever two different cultures come into sustained contact.This latter understanding of romanisation combines best the theoretical andempirical approaches to the subject and is the one that has been employedthroughout this study.Roman rule affected the native population, as well as the whole land-

scape in the conquered territories. The emergence of Roman-type towns,the broad diversification of the range and function of settlements, and theparticular way of organising space probably had a more significant impacton the pre-existing system than any other previous changes during prehis-tory. Comparison with other provinces of the empire can reveal particularaspects of this process of change, as well as indicating the real scale ofthe whole process within the territory. By combining archaeological andhistorical information, with information regarding the natural landscape, itis possible to understand better the general evolution of the landscape andthe human impact upon it, both in the pre-Roman and Roman periods.In such a context, it should be possible to distinguish from a more realisticstandpoint, given the amount of data available, exactly what constitutes thegeneral pattern and what can be considered unusual; to identify evolutionarypatterns; and to consider the occurrence of special cases and their possiblecauses.The purpose of my research is to redress some of the bias that hinders

current interpretations of how Dacia became Roman. The nature of currentresearch on rural Roman Dacia described above significantly biases the evid-ence for any such analysis of the economic and social life of the province.Since it has not been recognised before, this bias raises serious doubts aboutthe validity of currently accepted theories about the development of thislandscape. The potential density of human settlement in the period has notbeen fully appreciated and the typology of rural sites might not be complete.Accordingly, the native pre-Roman component in the life of the province

Introduction 19

has probably been misinterpreted. All these issues affect the evaluation ofthe Roman impact on the conquered territory and the nature of the roman-isation process in Dacia. The study is centred on the effects of the Romanoccupation on the indigenous settlement pattern and land-use. From aninterpretative point of view, there are several questions to be addressed. Inwhat way did the Roman conquest affect the native landscape? What werethe mechanisms behind the choice of settlement location and which of thefactors influencing decisions are predominant in the case of different types ofsettlements in Roman Dacia? Can we detect the evidence to support the ideaof a state-directed policy of settlement emergence and pattern in the caseof Roman Dacia, as has recently been suggested, or is the impact of theRoman colonists the product of multiple small-scale individual strategies?Did the conquest result in any perceptible resistance phenomena amongstthe natives? Finally, how did the process of romanisation develop in Dacia?Approaching these questions will offer the opportunity to address a numberof current debates and assumptions: whether archaeological evidence bearsout the literary references to depopulation, whether much of the hinterlandof Sarmizegetusa was unoccupied in the pre-Roman period, or whether landwas parcelled out and given to the colonists. The understanding of the realRoman impact, military and civilian, and of the true nature of the socialrelationship established between the conquerors and natives will ultimatelylead to a better understanding of romanisation in Dacia.The present study does not cover the whole of Dacia, but focuses on the

area within the territory surrounded by the Carpathian Mountains whichwas the geographical core of both pre-Roman and Roman Dacia. It includesthe colonial settlement and provincial capital at Sarmizegetusa (Colonia UlpiaTraiana Sarmizegetusa) (Figure 1.4) along with its hinterland in the area ofHateg (Tara Hategului) and the Strei River valley, extending further alongthe whole middle Mures River valley, beyond the colony and legionary baseat Alba Iulia (Apulum) (Figure 1.5). The choice of looking in particular atthe lowlands of Tara Hategului and mid-Mures valley, though includingthose parts of their surrounding uplands that define them and are structurallyrelated (Figure 2.1), is deliberate. More extensive upland areas taken intoconsideration are the Orastie Mountains, as the core of Iron Age kingdomof Dacia (Figure 1.6), and the Roman iron-mining district from the PoianaRusca Mountains (included in the study in order to balance the pre-Romanfocus of iron extraction in the Orastie Mountains). But on the basis ofits scale and exclusive focus on mining, the most extensive mining areaof Roman Dacia (the gold-mining district from the Apuseni Mountains,located to the north of the mid-Mures valley) can be considered a highlyspecialised landscape in its own right; not readily comparable with either thelowlands or the uplands included here, it was deliberately excluded in orderto eliminate potential bias.Given that the traditional approach is so much out of date, especially when

it comes to rural sites, the analytical perspective focuses on the evolution

20 Introduction

Figure 1.4 Ulpia Traiana Sarmizegetusa – general view of the Roman town andmodern village from the east, with the Iron Gates of Transylvania inthe background.

of the settlement pattern within the landscape. The quantity and qualityof information available on particular sites is hardly satisfactory, and theproblems in accessing it have made this endeavour extremely difficult. Asmany as 60 per cent of the archaeological sites considered here are merelyaccidental discoveries, mainly in the form of artefacts, or their mode ofdiscovery is not mentioned in publication; a further 10 per cent come fromantiquarian reports. Some 17 per cent have been subject to excavation atvarying scales and a further 8 per cent have been reported through fieldwalking, although many cannot be accurately located.In addition, new information has recently become available through

modern methods of archaeological prospection, particularly aerial recon-naissance which has been shown to provide some of the best resultsfor landscape studies (e.g. Palmer 1984; Stoertz 1997). My researchmakes primary use of photographic information produced by the GlasgowUniversity aerial reconnaissance programme in western Transylvania, whichcovered the same area (Figure 1.3). The nature of the local soils,mainly alluvial clays, which are heavier and retain moisture better thansandy soils, favours the formation of negative cropmarks indicating the

Introduction 21

Figure 1.5 Apulum – general view from the south of the site of the Colonia Aurelia,now the Partos suburb of Alba Iulia; the eastern, northern and the earlywestern sides of the enclosure are visible on the edges of the modernoccupation.

presence of buried roads, ramparts or stone buildings common on sites ofRoman date (Hanson and Oltean 2003; Hanson 2005b; Oltean and Hansonforthcoming c). Unfortunately, positive cropmarks representing ditches, pits,drains or sunken-houses common to agricultural villages and pre-Romansettlements are less frequently revealed. This means that the recovery ofnative-type settlements by aerial photography has been significantly reduced,creating a potential bias that needs to be taken into consideration. In addi-tion, as demonstrated at Micia (Oltean et al. 2005) this bias in the dataprovided by aerial reconnaissance means that the cropmark evidence reflectsthe stone phases of construction at the settlements, for only in the verybest conditions are the slight traces of the construction trenches of timberbuildings visible as positive cropmarks. In addition, only rarely can crop-marks indicate different phases in stone constructions on the basis of differ-ences in alignment and overlapping features (e.g. Cigmau, Oarda, probablyRazboieni – see Chapter 5). Given these problems, only some 5 per centof the sites considered here represent new or augmented discoveries fromthe air, but the qualitative balance of the different sets of data is effectivelyreversed.

22 Introduction

Figure 1.6 General view of the Gradistea Muncelului hill, the site of the Daciancapital, Sarmizegetusa Regia.

In order to produce general conclusions, all the information has beenbrought together into a coherent system to facilitate both area-wide anddetailed on-site analysis, and permit its evaluation as a whole. In parallel, theevidence that formed the fragile basis of previous theories has been reviewedin order to see how much of what is currently asserted is, indeed, based onfact and how much which started merely as hypothesis ended up as acceptedfact. Large amounts of published data, collated from various reports of occa-sional field walking, the existing gazetteers, excavation reports and otherpublications, not only from the twentieth century, but also of earlier date,was integrated with the new data provided from aerial reconnaissance. Thiswas augmented with archaeological and landscape transformation inform-ation contained in a number of archived representations of the area. TheTabula Peutingeriana offers a unique account of settlement evidence fromDacia that has previously been considered, but the correlation with archae-ological evidence and especially with the location of these sites has providedthe overall analysis with additional grounds for interpretation of the func-tions of these sites and their place within the landscape. Several sheets of the1:28,800 Austrian cadastral survey of Transylvania (1870–1875) containedvaluable archaeological information (Figures 1.7–1.9) that was integratedwith other data. But apart from direct information on archaeological sites,

Figure 1.7 Remains of Roman sites as represented on the 1870–1875 Austriancadastral map of Transylvania: colonia Sarmizegetusa and itssurroundings (The Austrian State Archives, Vienna).

Figure 1.8 The auxiliary fort at Micia as represented on the 1870–1875 Austriancadastral map of Transylvania (The Austrian State Archives, Vienna).

24 Introduction

Figure 1.9 Line of the Roman road along the Mures from the river crossing nearGelmar, towards Sibot (The Austrian State Archives, Vienna).

archive mapping along with early aerial imagery provide information on theevolution of the landscape prior to extensive modern development. There-fore, various editions of modern maps were compared with the latest editionsavailable. Unfortunately, Romanian archived aerial photographs could notbe consulted, but a limited set of high-altitude aerial photographs from theSecondWorldWar covering areas in the north-eastern part of Tara Hateguluiand along the Mures valley around Simeria and Teius could be locatedin the Aerial Reconaissance Archive in Keele University. In addition, first-generation satellite imagery (CORONA) declassified by the United Statesin 1995 was utilised along the Mures valley between Zam and Tartaria. Allthis historical mapping and imagery offered the opportunity to evaluate thelandscape before it was affected by the later development. Finally, the lastset of archived data utilised is very recent (27 March 2003) and consists oflarge resolution QuickBird satellite imagery for areas around Alba Iulia andAiud, freely available on the Internet through Google Earth; although oflimited coverage, it has made an important contribution in documenting thelandscape at its current stage of development in a different season than thatnormally employed for aerial reconnaissance (June to July) and in assistingthe process of transcription and mapping of oblique aerial photographs.This book offers first a description of the natural environment of western

Transylvania, its topographic setting, climate, resources and environmentalchanges from late antiquity to modern times (Chapter 2), followed by aconcise but comprehensive historical framework of the Dacian area andits conquest and organisation as a province by Rome (Chapter 3). Thereal subject of the book is the people living in the study area and the

Introduction 25

transformation of the landscape. The framework is that of a profoundly socialand active landscape. Society created this landscape for its own convenienceand this is expressed through the types of settlement, the choice of theirlocation and in the way main activities were carried out therein: work and useof available resources, worship and death, networking and administration.In presenting the overall picture of settlement and land-use in central Daciaduring pre-Roman and Roman times, Chapters 4 and 5 are intended tomirror each other. The aim of this parallel presentation is to offer the readerthe opportunity to better understand and form an opinion on the similaritiesand the differences, and the pattern of continuity and disruption visible in thedataset presented. Finally, Chapter 6 discusses in detail the way in which theRoman conquest affected the native landscape following the transformationsin the settlement hierarchy, typology and choice of settlement location, andthe specific impact of the Roman army on the creation of this new landscape.It will try to give an answer to some of the preliminary questions raisedabove and offer an alternative interpretative scheme for the way the processof romanisation developed in Dacia.

2 The Dacian heartland

2.1 Physical geography: geomorphology, topography and geology

Modern Romania is located in Eastern Europe, to the north of the BalkanPeninsula and west of the Black Sea. With an even distribution of moun-tains, hills and plains and a rich network of watercourses, the geography ofRomania is largely structured around the south-eastern end of the CarpathianMountains (a branch of the Alpine–Himalayan Mountains). The curvingline of the mountains imposes a concentric layout to the general geographyof the country and outside the mountains the sub-Carpathian hills and theplains spread out in steps. Inside, however, the mountains enclose a large areaof hills, tableland and alluvial plains, called Transylvania. The name itselffirst occurs in early medieval Hungarian chronicles of the eleventh centurywritten in Latin (Anonymus, Simon de Keza) as the land ‘beyond the forests’(trans silvae) (Pop 1997, 5–7) that once covered much of the Carpath-ians. Transylvania can be understood as a space enclosed by the mountains.This topographic characteristic has determined various interpretations of theadvantages that the area has offered to human settlement throughout history.Opinions vary from ‘citadel’ to ‘meeting point’, apparently in contradiction,but it is exactly the particularity of its topographic and geographical settingthat makes both interpretations equally true.The Carpathians surrounding Transylvania were formed in the post-

Mezo-Cretaceous and are characterised by medium and low altitudes, whichaverage 1,000metres, with valleys of around500metres in depth.Thesemoun-tains are very fragmented, both longitudinally and transversally, by numer-ous depressions and river valleys, making them easier to cross from one sideto another. Some of the mountains are of younger, volcanic origin, but mostof them were created by the folding movements that happened at the end ofPliocene and the beginning of the Quaternary period (Gherasimov 1960, I,197). Affected by these movements, Transylvania first slowly sank and wasin-filled by marine and continental deposits of up to 4,000 metres in thick-ness and was transformed into a large plain. Later on, rising movements atthe beginning of the Quaternary transformed most of it into a hilly regiondefined by the piedmonts and internal sub-Carpathian Hills located at the

The Dacian heartland 27

contact area with the mountains and the Transylvanian Tableland in themiddle (Gherasimov 1960, I, 197). The water from the interior drainedaway through the main river valleys, though some of the ‘gulfs’ locatedat the contact area with the mountains remained under water until muchlater (Morariu et al. 1966, 27), when they became depressions such as TaraHategului.The western side of Transylvania (Figure 2.1) includes the mid-Mures

valley between Ocna Mures – Razboieni to the north and Zam-Savarsinto the west, and the whole Strei River valley and the Hateg depression tothe south. The area is surrounded by higher grounds rising gradually onboth sides of the valleys as terraced sides of the internal sub-CarpathianHills and the Western and Meridional Carpathians to the west and south,and the Transylvanian Tableland to the east. This gives an amphitheatre-like appearance to the whole area, centred along the valleys of Mures andStrei.

Figure 2.1 Map of the main topographical features and the location of modernsettlements mentioned in text.

28 The Dacian heartland

The outer branches of the Apuseni Mountains border the Mures valley onits northern and north-western side. These mountains are of low altitude,ranging between 800 and 1,200 metres (Metaliferi and Trascau Mountains;Gheorghiu 2001, 2) and 400–600 metres (Zarand Mountains; Morariuet al. 1966, 25). With a particular mosaic of various rocks – sandstone,limestone and volcanic rocks, such as dacite, andesite and basalt – theMetaliferi Mountains are best known, as their name implies, for theirmetal ores, particularly gold (Floca 1957, 16). To the south the PoianaRusca Mountains are delimited by the Apuseni Mountains and the RetezatMountains, by the Mures Defile and by the passage obligée of the IronGate of Transylvania (Gherasimov 1960, I, 218). With a geology of meta-morphic schists (Floca 1957, 15), they have extremely rich resources of ironaround Hunedoara and copper and andesite in the vicinity of Deva. TheRetezat Mountains (2,518 metres) are amongst the highest in Romania.Also formed by schists and micaceous schists (Gherasimov 1960, I, 212),with only limited amounts of limestone in some of the valleys, their uppersides were strongly shaped by glaciers, whose remains can still be seenas multiple glacier lakes and numerous sources of deep and steep glaciervalleys (sometimes more than 1,000 metres in depth) with watercourses thatdescend in steps towards the lowlands (Gheorghiu 2001, 3). The SureanuMountains (also called ‘Sebes’ or ‘Orastie’ Mountains) are lower than theRetezat (2,130 at Virful lui Patru). Their upper parts are fairly level ondifferent steps without much fragmentation. With a higher centre (1,600–1,800 metres), their outer limits to the north in Transylvania reach only900–1,100 metres (Gheorghiu 2001, 3). Their flattened appearance is evenmore evident at their western end, which extends into a large elevatedplatform (the Luncanilor Platform). Located at the south-western end, atthe point of contact with the Hateg Depression, is a large carstic zone(Ohaba Ponor-Banita) that was formed on a base of Jurassic limestones withmany caves, dolines, canyons and sub-terranean rivers (Gherasimov 1960,I, 216).The geological background of the Transylvanian Tableland is represented

by clays, marls and sand, with limestone and volcanic intrusions. The easternhalf, where the aspect is of high hills and plateaux fragmented by rivervalleys, is of higher altitude than the western side and in some places reacheseven 600–700 metres in height (Morariu et al. 1966, 32). The south-westernarea (also called the Secaselor Tableland) is an area of monocline slopingand small depressions, along with areas of ongoing erosion, while the hillsfurther north along the Mures River (450–500 metres) have broad archesand terraces along the river valleys (Gherasimov 1960, 230–1).Hills with smooth slopes regularly dissected by watercourses, grouped

around river terraces and valleys are also located at the point of contactbetween the mountains and the alluvial plain. The hills bordering the moun-tains have a general aspect of multiple piedmontaneous steps and riverterraces. To the south of the Metaliferi Mountains, the narrow area between

The Dacian heartland 29

the mountains and the Mures River is occupied by hills mainly made ofslate, but volcanic stone (augite–andesite) is found localised at Uroi, northof Simeria, and around Deva (Floca 1957, 16). Large quantities of alluviumresulting from denudations created by uplifting movements of the moun-tains during the Pliocene and Quaternary were transported into the valleyscreating piedmontaneous plains in the southern and eastern side of the HategDepression and in the hills from Orastie to Sebes (Gherasimov 1960, 212).The depression around Hateg (Tara Hategului) is in fact a piedmont plainwith fan-shaped terraces and cones of alluvium arranged in three concentricsteps, and with dense watercourses.The plains occupy a relatively limited area. All of them are alluvial in origin

and were developed along the Mures and its main tributaries, the Aries,Tarnava (with the Tarnava Mare and Tarnava Mica), Ampoi, Sebes and Strei(with Rau Mare and Galbena). Other tributaries, like the Cugir, Orastie,Geoagiu or Cerna, have produced smaller impacts in terms of topographyand outflow. Many other watercourses in the area are nothing more thanstreams. The river Mures originated in the eastern Carpathians and, withits length of 880 kilometres and outflow average of 70 cubic metres persecond (Floca 1957, 20), is regarded as the most important tributary of theTisa River (Morariu et al. 1966, 46). Its course is generally oriented E–W,though its medial segment follows the contact line between the internalsub-Carpathian Hills of the western Carpathians and the TransylvanianTableland. The Mures is a very active river; its alluvial deposits have createda large fertile valley up to 8 kilometres wide. The general appearance ofthe valleys formed by its tributaries is of corridors of variable width thatincrease in their lower courses immediately after they exit the mountains.They make an important contribution both to the general outflow of theMures and to the quantity of alluvium it carries, so that the plain of theMures is generally wider at its confluence with tributaries, providing spacefor agriculture and human settlement. The tributaries are also responsiblefor the changes of direction of the main course of the river and enforcedthe creation of multiple meanders (Figure 2.2). The meadow land along theMures, Strei, lower Sebes, Cerna and Orastie valleys, and the mid- and lowervalley of the Ampoi, sometimes come under threat of flooding (Gheorghiu2001, 5). Along the valleys fairly parallel terraces were developed, usuallyfrom six to seven in number, but up to eight at the contact zone with thehills and tablelands; they generally have a horizontal aspect (Gherasimov1960 I, Table 1), but around the Tarnave region and in the area betweenSebes-Vintu de Jos and Deva their slight downward sloping increases thelevel of erosion.In the plains the soil cover is composed of alluvial soils, alluvial proto-

soils and chernosem, in some places still retaining salt deposits fromthe draining of the sea long ago. The soil of the mountain, hill andtableland regions is mainly represented by a large variety of forest soilsaffected by erosion (podsolisation) to various degrees. There are also

30 The Dacian heartland

Figure 2.2 River meanders in the Strei valley near Calan.

areas of chernosem, developed above clays and marls along the lowerSebes valley and to the south of the Mures valley between Sebes andSimeria, in the vicinity of Alba Iulia and Teius and around Razboieniand Deva, which also correspond with the confluence zones of the maintributaries (Aries, Tarnava, Sebes, Strei). Not surprisingly, these partic-ular areas are also known to be the most productive in terms of cerealcultivation.From several points of view, the study area does not constitute a unitar